AI in theory

So, what are we talking about when we talk about AI?

Essentially, it’s technology that can simulate or mimic human intelligence in machines or computer systems. This enables technology to perform tasks that have typically required human intelligence, such as problem-solving, reasoning, learning, perception and language understanding.

Large language models such as ChatGPT, which build and train deep neural networks with multiple layers of interconnected artificial neurons to process and learn from vast amounts of data, have made a particularly big impression on the public consciousness.

But AI is only just breaking into the business world, and with the possibilities seemingly endless, it can be hard to determine best practice usage.

Brent Wallace, co-founder and principal responsible for product strategy at Hatch Head, a multi-disciplinary agency focusing on software design and user experience, has found AI to be particularly helpful for translating unstructured data types – such as text files, images, social media data, emails and audio and video files – into a structure that a machine can use.

However, he also acknowledged that having an authentic human voice is always going to be incredibly valuable.

“AI is more artificial than intelligence, and should be used in a considerate and invisible manner,” he told the attendees.

Matthew Clarkson, CTO of Cemoh, has recently been working on integrating AI into his company’s platform, as well as building a number of other AI tools in the marketing space.

“AI is currently as bad as it is going to be,” he said. “It can only get better from here.

“It exponentially increases our capacity to solve problems, whether it be for mass production of content or paid data analysis and interpretation.

“It increases our effectiveness and gives us the ability to create tailored solutions for our businesses.”

 

AI in practice

When utilising AI in the workplace, both Brent and Matthew agreed that AI can get the grunt work done, but the need for humans still remains.

“We should be treating AI like our intern,” Matthew said. “It gets the first 80 per cent of the work done and the rest is human.

“It shouldn’t replace humans, but it should make us more productive and efficient in our jobs.

“There are lots of tools in the world that have changed our lives, rather than taking our jobs. They’ve made us more productive, more efficient, have improved our quality of life and provided a higher life expectancy.”

Brent agreed that the benefits of AI could be phenomenal.

“If we utilise AI effectively, we could be looking at reducing our work week or being more productive within the working day,” he said.

“We can’t get caught up in the flashy, speculative side of things – instead we should be looking at opportunities to use technology to get things done and solve real problems.”

Brent pointed to Hatch Head’s use of AI to assist with logistics, and the positive results this has already yielded.

“For the longest time, unstructured data types have been the bane of my existence when it comes to programming and coding,” he said.

“Coordinating international shipments is incredibly convoluted and hard, involving a huge number of people communicating back and forth on a variety of email chains, with that data then being transferred into large volumes of complex spreadsheets that need to be managed.

“It’s a trillion dollar industry that is growing six per cent every year and runs entirely on email and outdated systems. It works, meaning there isn’t a lot of appetite to disrupt it.

“But with AI, we are able to take the unstructured information in emails, and in less than a second we’re able to put it into a format that allows the system we’re building to actually interpret and take the information in.

“We don’t want to move away from email, because it’s a system that works. Instead, we want to embrace it while making the experience a lot easier.

“While we are still in the early stages, our end goal is to be able to receive an email, hit a button, and for that email to be transferred into a data set that the system can use straight away.”

For Matthew, the productivity and efficiency benefits of Cemoh’s use of AI have also been great, especially when it comes to producing content.

“We now have the ability to generate our own unique stock imagery and videos for a fraction of the price it is usually available for,” he said.

“Many of the graphic designers and videographers in our company are using these tools to improve their workflows, freeing up time for additional tasks.

“We are also using it in written content to produce things like transcripts, blogs and emails, and for summarising bodies of work and doing procedural work.

“The results are 10 times the content you could have produced yourself, with 10 times the efficiency in the production.

“So we’ve got the ability now to produce a lot more content, a lot more efficiently.”

 

The limits of intelligence

Not wanting to leave attendees with the impression that AI could be the magic bullet to solve all their problems, our speakers also warned of the technology’s limitations.

“We are definitely seeing a lot of what we call ‘solutionism’,” Brent said. “This is when we go looking for problems we can apply technology to when it is entirely unnecessary.

“We’ve seen examples of using Bluetooth on your toothbrush and AI in your fridge. It just isn’t necessary.

“Instead, we should be considerate about using it to solve real world problems.”

Matthew added he has seen some troubling bias in training data in his work with AI, particularly with regards to image generation.

“The AI models are only good at generating what they are accustomed to seeing,” he said. “This is leading to poor diversity representation and stereotyping and an inaccurate representation of body types in AI-generated images.

“While these problems will improve over time, it is something to be aware of when using the platforms.”

Matthew also stressed that the legal issues around AI are far from settled.

“Everything you upload into these AI models becomes open source for the world to use, meaning we have to be incredibly careful with what we are inputting into these platforms,” he said.

“We’re also seeing issues around copyright and trademark infringement. It’s really important to review and customise AI-generated work so it isn’t a direct imitation.”

Matthew noted the increasing ubiquity of AI also leads to the more ready availability of illegal imagery and deep fakes – media that’s been digitally manipulated to convincingly replace one person’s likeness with another.

 

What’s next?

Brent noted that we’re already seeing an explosion in platforms that are integrating AI into their systems, but we’re still at the beginning of the curve.

“I think every single platform is going to have some sort of AI part to it, for better or for worse,” he said.

“Before we perfect the usage, I think we’re going to go too far with it and there’s going to be a lot of rubbish produced. But the companies that nail it and are left standing will be the ones that solve real problems.

“It’s still the wild west out here, in terms of how AI is going to be used, and I’m looking forward to seeing what is to come.

“At the end of the day, I predict having a human and authentic voice is going to be even more valuable, because most people will rely on AI.”

Matthew said he ultimately sees AI as a tool for supercharging human work, but warned that we’re going to see a lot of bad work in the meantime.

“I think it’s going to take time to master,” he said. “I would never recommend blindly publishing AI-generated content, but instead using it to provide outlines and get inspiration that we as humans can then curate.

“Inputting your own content into these AI language models also leads to personalised outputs that are unique to you and your company, so I think the opportunity for customised AI is going to be huge in the next six to 12 months.

“It’s going to be a game changer.”

To learn more about using ChatGPT and automation in your marketing strategy, read this article with OH! Marketing owner and director Michelle O’Hara.